CENG17 Coursework 1

1- Dynamic model
The model takes into account the following assumptions:
a) CSTR achieves perfect mixing within its volume.
b) F 0=F (given)
c) No biomass in the feed hence x 10 =0 (given)
d) A constant liquid volume (V) within the reactor is maintained by an overflow mechanism, ensuring that the dilution rate (D)
remains constant as well.
e) The mass densities (ρ) of both the feed and effluent streams are constant and equivalent.
Using the general mass balance equation, we can determine the mass balances for the biomass and the substrate:
Accumulation = Input – Output + Generation
Starting with the biomass, its rate of change in the reactor is determined by the net effect of biomass outflow and the biomass
generated within, expressed as:
d ( V x1 )
=F 0 x 10−F x 1 +V r 1
dt
( 1)
where V is the CSTR volume (L), x 10 and x 1are the inlet and outlet biomass concentration (g/L) respectively, F 0 and F are the inlet
and outlet volumetric flowrate (L/hr) respectively, and r 1 is the rate of biomass generation (g/ L ∙hr ¿ .
Considering assumptions b), c) and d) equation 1.1 can be simplified to:
d ( x1 )
V =−F x 1 +V r 1
dt
( 2)
F
Since the dilution rate D is expressed as D= , Eq 2 can be rewritten, after dividing by V, as:
V
d ( x1 )
=r 1−D x 1
dt
(3 )
d ( x1 )
The rate of change of biomass concentration over time is described by the equation:
dt
d ( x1 )
=μ ( x 2 ) x1−D x 1
dt
(4 )
where μ ( x 2 ) x 1=r 1represents the biomass generation rate, μ ( x 2 )being the specific growth rate dependent on the substrate
concentration in the effluent, measured in 1/hr. By considering two different models for μ ( x 2 ), we obtain Eq 5 for the Monod Model
and Eq 6 for the Substrate Inhibition Model:
μm x2 x1 μm x 2 x 1
μ ( x 2 )= (5) and μ ( x 2 )= 2
(6)
k 1+ x 2 k 1 + x 2+ k 2 x 2
Finally, equations 5 and 6 can be substituted into Eq 4 to obtain the biomass mass balance in relation to the Monod Model (Eq 7) and
the Substrate Inhibition Model (Eq 8):
d ( x1 ) μm x 2 x 1 d ( x1 ) μm x 2 x 1
= −D x 1 (7) and = 2
−D x 1 (8)
dt k 1+ x 2 dt k 1 + x 2 +k 2 x2
While the rate of change of the substrate in the reactor is characterized by the net input and output of substrate minus the substrate
consumption all within the CSTR expressed as:
d ( V x2 )
=F 0 x 20−F x 2−V r 2
dt
( 9)
where 20 and 2 are the inlet and outlet substrate concentration (g/L) respectively, and r 2 is the rate of substrate consumption (g/
x x
L ∙hr ¿.Considering assumptions b) and d), Eq 9 simplifies to:
1
 d ( x2)
V =F ( x 20−x 2 )−V r 2
dt
(10 )
F
Using the dilution rate D= , and dividing through by V, the substrate balance equation (Eq 10) becomes:
V
d ( x2 )
=D x 20−D x2 −r 2
dt
( 11 )
r1
To incorporate the biomass yield defined as Y = and since r 1=μ ( x 2 ) x 1, we can express r 2 in terms of r 1 and consequently as
r2
μ ( x2 ) x1 d ( x2 ) 1
. Substituting into Eq 11 we get: =D x 20−D x2 − μ ( x 2 ) x 1(12)
Y dt Y
Incorporating the Monod and Substrate model relationships for μ into Eq 12, we derive the substrate mass balances for both the
Monod model (Eq 13) and Substrate Inhibition model (Eq 14):
d ( x2 ) 1 μm x 2 x 1 d ( x2 ) 1 μ m x 2 x1
=D x 20−D x2 − ( 13 ) and =D x 20−D x2 − (14)
dt Y k1 + x2 dt Y k 1+ x 2 +k 2 x 22
2- Linearized model
To start, we consider the biomass’s nonlinear differential equation as per the Monod Model (Eq 7):
d x1 μm x 2 x 1
=f 1 ( x 1 , x 2 )= −D x 1
dt k 1+ x2
( 15 )
Linearization involves expanding Eq 15 using a Taylor series at the steady-state points (x ¿ ¿ 1 s , x 2 s )¿ , and retaining only the linear
terms as follows:
d x1 ∂f1 ∂f1
=f 1 ( x 1 , x 2 ) ≈ f 1 ( x 1 s , x 2 s ) + ( x1 s , x 2 s ) ( x1 −x1 s) + ( x , x ) ( x −x )
dt ∂ x1 ∂ x2 1s 2s 2 2s
(16 )
μm x2 s x1s
since f 1 ( x 1 s , x2 s ) = −D x 1 s substituting into Eq 16 yields:
k1 + x2 s
d x1
dt
=f ( x 1 , x 2 ) ≈
1
μm x 2 s x 1 s
k 1+ x 2 s
−D x 1 s + (
∂ f 1 μm x 2 s x 1 s
∂ x 1 k 1+ x 2 s )
−D x 1 s ( x 1−x 1 s ) + ( ∂ f 1 μm x2 s x1s
∂ x2 k 1 + x 2 s )
−D x 1 s ( x2 −x2 s)

( 17 )
Then eq 17 underwent partial differentiation, which was then simplified to get the linearized equation:
d x 1 μm x2 s (k1 μm x1 s ) ( k1 μm x1 s )
= x 1−D x 1 + x−
2 2
x
2 2s
dt k 1 + x 2 s ( k1 + x2 s ) ( k 1+ x 2 s )
( 18 )
d x1s μm x 2 s x 1 s
We define at steady state that =f 1 s ( x1 s , x 2 s )= −D x1 s (19), hence subtracting Eq 19 from Eq 18, and introducing
dt k1 + x2 s
deviations ^x 1=( x 1−x 1 s ) and ^x 2=( x 2−x 2 s ), we arrive at the linearized model:
d x^ 1
dt
=
(
μm x2s
k 1+ x 2 s
−D x^ 1 +
) ( μm k1 x1s
( k1 + x2 s )
2
)
x^ 2

(20 )
Now, we will consider the substrate nonlinear differential equation for the Monod model (Eq 13):

2
 ( )
d x2 1 μ m x 2 x1
=f 2 ( x 1 , x 2 , x 20 )=D x 20−D x 2−
dt Y k1 + x2
(21 )
We linearize Eq 21 around the steady state points ( x 1 s , x 2 s , x 20 s ) by retaining only the first-order terms:
d x2 ∂f2 ∂f2
=f 2 ( x 1 , x 2 , x 20 ) ≈ f 2 ( x 1 s , x 2 s , x20 s ) + ( x 1 s , x 2 s , x 20 s ) ( x 1−x 1 s ) + ( x , x , x ) ( x −x )
dt ∂ x1 ∂ x 2 1 s 2 s 20 s 2 2 s
+∂f 2
(x , x , x )
∂ x20 1 s 2 s 20 s
( x 20−x 20 s ) (22)
since f 2 ( x 1 s , x2 s , x 20 s )=D x 20 s−D x 2 s−
( ) 1 μm x2 s x1s
Y k1 + x2 s
substituting into Eq 22 yields:

d x2
dt
=f 2 ( x 1 , x 2 , x 20 ) ≈ D x 20 s−D x 2 s− ( ) Y k 1+ x 2 s ∂ x 1 (
1 μm x 2 s x 1 s ∂ f 2
+ ( )
D x 20 s −D x 2 s−
)
1 μm x 2 s x1 s
Y k 1+ x 2 s
( (
x 1−x 1 s ) +
∂f2
∂ x2 (
D x 20 s−D x 2 s−
Y
Then eq 23 underwent partial differentiation, which was then simplified to get the linearized equation:

( ) ( )
d x2 1 μm x 2 s 1 μm k1 x1 s
dt
=−
Y k1 + x2 s
x 1−D x 2+ ( x −x ) + D x 20
Y ( k 1+ x2 s ) 2 2 s 2
( 24 )

( )
d x2s 1 μm x 2 s x 1 s
We define again at steady state that =f ( x1 s , x 2 s , x 20 s )=D x 20 s−D x 2 s− (25), hence subtracting eq 25
dt Y k1+ x2s
from eq 24 and introducing deviations ^x 1=( x 1−x 1 s ), ^x 2=( x 2−x 2 s ) and ^x 20 =( x 20−x 20 s ), we get the linearized model:
d x^ 2 −1 μm x 2 s
dt
=
Y k 1+ x 2 s ( ) (
^x 1+ −D−
1 μm k1 x1s
Y ( k 1+ x2 s )2 2
x^ + D ^x20
)
( 26 )
Therefore, the Monod Model can be written in the form of a matrix as d ^x /dt =A ^x + b ^x20 :

[ ][ ]
d ( x^ 1 ) μm x 2 s μm k 1 x 1 s
−D

[ ][ ]
dt k1 + x2 s ( k1 + x2 s )
2
x^ 1
= + 0 x^ 20
d ( ^x 2 ) −1 μm x2 s 1 μm k 1 x1 s x^ 2 D
−D−
dt Y k1 + x2 s Y ( k 1 + x 2 s )2
(27)
We will now proceed to the biomass's nonlinear differential equation related to the Substrate Inhibition model (Eq 8):
d x1 μm x2 x1
=f 1 ( x 1 , x 2 )= 2
−D x 1
dt k 1 + x 2 +k 2 x 2
( 28 )
Given its nonlinearity, we deploy a Taylor series expansion at the steady-state coordinates ( x 1 s , x 2 s ) keeping only the first linear term.
μm x 2 s x 1 s
Here, f 1 ( x 1 s , x2 s ) = 2
−D x 1 s, hence substituting into eq 28 gives the linearized form:
k 1 + x 2 s+ k 2 x2 s
d x1
dt
=f 1 ( x 1 , x 2 ) ≈−D x 1 s +
μm x2 s x 1 s
2
+
∂f1
k 1 + x 2 s+ k 2 x 2 s ∂ x 1 (
−D x1 s +
μ m x 2 s x1 s
2 ( 1
k 1 + x 2 s +k 2 x 2 s )
x −x 1 s ) +
∂f 1
∂ x2
−D x 1 s +
( μm x 2 s x1 s
2 ( 2
k 1 + x 2 s +k 2 x 2 s
x −x 2 s ) (
)
After performing partial differentiation on Eq 29 we will get:

3
 (
μm x1 s ( k 1−k 2 x 2 s )
)
2
d x1 μm x2 s
2 ( 2
= x 1 −D x 1 + x −x 2 s )
dt k 1 + x 2 s+ k 2 x 2 s2 ( k 1+ x 2 s + k 2 x 2 s 2 )
(30 )
d x1s μm x 2 s x 1 s
We define again at steady state that =f 1 s ( x1 s , x 2 s )= 2
−D x 1 s (31) hence subtracting Eq 31 from Eq 30 and
dt k 1 + x 2 s +k 2 x 2 s
introducing deviations ^x 1=( x 1−x 1 s ) and ^x 2=( x 2−x 2 s ) we will get the linearized form:

) ( )
μm x 1 s ( k 1−k 2 x 2 s )
(
2
d x^ 1 μm x2 s
= 2
−D x^ 1 + 2
x^ 2
dt k 1+ x 2 s + k 2 x 2 s ( k 1 + x 2 s+ k 2 x 22 s )
( 32 )
Now, we will consider the substrate nonlinear differential equation for the Substrate Inhibition model (Eq 14):

( )
d x2 1 μm x 2 x 1
=f 2 ( x 1 , x 2 , x 20 )=D x 20−D x 2−
dt Y k 1 + x 2+ k 2 x 22
( 33 )
Linearization involves expanding Eq 33 using a Taylor series at the steady-state points (x ¿ ¿ 1 s , x 2 s , x 20 s )¿ , and retaining only the

( ) 1 μm x 2 s x 1 s
linear terms. Here, f 2 ( x 1 s , x2 s , x 20 s )=D x 20 s−D x 2 s− , hence the linearized form will be:
Y k 1 + x 2 s +k 2 x 22 s
d x2
dt
=f 2 ( x 1 , x 2 , x 20 ) ≈ D x 20 s−D x 2 s− ( )1 μm x2 s x1 s
2
+
Y k 1+ x 2 s + k 2 x 2 s ∂ x 1
∂f2
( D x 20 s−D x 2 s−
1
( ) μ m x 2 s x1 s
)
2 ( 1
Y k 1+ x 2 s +k 2 x 2 s
x −x 1 s ) +
(
∂f 2
∂ x2
D x 20 s −

After performing partial differentiation on Eq 34 we will get:

( ) (
−1 μm x 1 s ( k 1−k 2 x 2 s )
)
2
d x 2 −1 μm x 2 s
2 ( 1) ( x 2−x 2 s ) + D x 20
= x −D x2 +
dt Y k 1+ x 2 s +k 2 x 2 s Y ( k + x +k x 2) 2 We know that at steady state
1 2s 2 2s

( 35 )

( )
d x2s 1 μm x 2 s x 1 s
=f ( x1 s , x 2 s , x 20 s )=D x 20 s−D x 2 s− (25), hence subtracting Eq 25 from Eq 35 and introducing deviations
dt Y k1+ x2s
^x 1=( x 1−x 1 s ), ^x 2=( x 2−x 2 s ) and ^x 20=( x 20−x 20 s ), we get the linearized model:

( ) ( 1 μm x1 s ( k 1−k 2 x 2 s )
)
2
d x^ 2 −1 μm x 2 s
= ^
x 1 + −D− ^x + D x^ 20
dt Y k 1+ x 2 s +k 2 x 2 s2 Y ( k + x + k x 2 )2 2
1 2s 2 2s

( 36 )
Therefore, the Substrate Inhibition model can be written in the form of a matrix as d ^x /dt =A ^x + b ^x20 :

[ ][ ( ]
μm x1 s ( k 1−k 2 x 2 s )
2
d ( x^ 1 ) μm x2 s
−D

[ ][ ]
dt k 1 + x 2 s +k 2 x 22 s (k 1+ x 2 s + k 2 x 22 s )
2
x^ 1 0 ^
= + x 20
d ( ^x 2 ) 1 μm x 1 s ( k 1−k 2 x 2 s ) ^x 2 D
)
2
−1 μm x 2 s
dt −D−
Y k 1 + x 2 s+ k 2 x 22 s Y ( k + x + k x 2 )2
1 2s 2 2s

( 37 )
3- Steady-state solutions (gPROMS)
In the context of the models under consideration, x 1and x 2 denote the concentrations of biomass and substrate, respectively, at a given
time. At steady state, the system's concentrations are indicated as x 1 sand x 2 s. To ascertain these equilibrium concentrations, we equate
the time derivatives in the mass balance equations to zero since at equilibrium the concentrations' rate of change over time is zero. The
mass balances for the biomass and substrate in terms of the Monod model (Eq 7 and 13 respectively) would change to:

4
 ( )
μm x2 x1 1 μm x2 x 1
0= −D x1 (38) and 0=D x 20−D x 2− (39)
k1+ x2 Y k 1+ x 2
Identically, the mass balances for the biomass and substrate in terms of the Substrate model (Eq 8 and 14 respectively) would turn to:

( )
μm x2 x 1 1 μm x 2 x 1
0= 2
−D x 1 ( 40) and 0=D x 20−D x 2− (41)
k 1+ x2 + k 2 x 2
Y k 1 + x 2+ k 2 x22
Then, the biomass and substrate are modeled on gPROMS, in terms of the Monod model using Eq 38 and 39, as well as in terms of the
Substrate Inhibition model using Eq 40 and 41. The resulting steady-state concentrations are depicted in table 1. In contrast to one
stationary solution for the Monod model, two stationary solutions for the biomass and substrate concentrations were found with regard
to the Substrate Inhibition model, as shown in Table 1. When the default value was adjusted to 0.5, the trivial solution of x 1 s= 0 g/L
and x 2 s= 4 g/L was achieved, while setting it to 0.6 resulted in the non-trivial solution.
Table 1: Steady state concentrations of the biomass and substrate for both models
Parameters Units Monod model values Substrate Inhibition model values
−1
Concentration of biomass (x ¿¿ 1 s)¿ g.L 1.53739 1.53016 0
Concentration of substrate ( x 2 s ¿ g.L
−1
0.156522 0.174592 4

4- Dynamic simulations (gPROMS)

1. Nonlinear model in terms of concentration for each cases of both models

The Figure 2: Plot representing the concentration of substrate vs time for Monod model
Figure 1: Plot representing the concentration of biomass vs time for Monod model

consumption of substrate by biomass is demonstrated by the gradual decrease in substrate concentration x 2 and increase in biomass
concentration x 1in Figures 1 and 2. Following the previously determined steady-state solutions, cases 1 and 2 both yield to steady
state with x 1 sat 1.53739 g/L and x 2 s at 0.15652 g/L. In case 1, it takes 35 hours for x 1and 44 hours for x 2 to achieve a steady state,
while in Case 2, it reaches a faster steady state—27 hours for x 1and 30 hours for x 2—because the starting biomass concentration is
closer to its ultimate value. Following a hyperbolic pattern, the growth rate increases with substrate concentration before stabilising at
the maximum rate.

Different behaviors in the Substrate Inhibition model are shown in Figures

3 and 4. As predicted by the Monod Model, the system reaches a steady
state in case 5, which replicates the initial conditions of case 2. The
biomass concentration x 1is 1.53016 g/L, while that of substrate x 2is
0.174593 g/L. The system stabilises at 27 and 30 hours, respectively. In
contrast, and contrary to predictions, cases 3 and 4 show a decrease in x 1
and an increase in x 2. This is described by the growth rate equation μ as a
function of x 2, which is linked to the Substrate Inhibition model's

Figure 3: Plot representing the concentration of biomass vs time for Substrate Figure 4: Plot representing the concentration of substrate vs time for Substrate
Inhibition model Inhibition model

recognition of inhibitory effects at high x 2levels that lower growth rates.

5
Elevated x 2values considerably raise the denominator of this equation, lowering μ and x 1as a result. With x 1 reaching 0 g/L and x 2
rising to 4 g/L, cases 3 and 4 therefore fail to attain a steady state, suggesting an unstable steady state scenario under these
circumstances.
The Monod Model's eigenvalues of -0.3 and -3.1968 and the Substrate Inhibition model's eigenvalues of -0.3 and -2.2620 indicate that
any disturbances will ultimately decrease, according to the eigenvalue analysis of the linear models, which suggests stability.
However, the non-linear models depict a more complicated practice. The linearized version of the model does not account for the non-
linear model's prediction that steady-state conditions may not be reached, especially in cases 3 and 4. The non-linear model offers a
more thorough illustration of how particular input variables and initial conditions can greatly affect the equilibrium of the biological
system, even though the negative eigenvalues mathematically indicate stability.
2. Linearized model for case 1 using both the Monod and the Substrate Inhibition models
With respect to small fluctuations around
equilibrium, the linearized model
provides an accurate approximation of steady-
state conditions under certain
conditions, as demonstrated by the
deviation term. As a result, the system
achieves the steady-state biomass and
substrate concentrations of 1.5379 g/L and
0.15622 g/L—quicker than the non- linear
model—in shorter amounts of time, 30 and
33 hours, respectively. But as the
departure from the equilibrium
increases, the reliability of this model
decreases, underestimating the
Figure 6: Plot representing the concentration of substrate vs time for both models
Figure 5: Plot representing the concentration of biomass vs time for both models (case 1)
(case 1)
duration to stability. Case 1 illustrates a particularly striking rapid shift to steady-state within the first hours, which may not accurately
reflect system dynamics because the initial conditions were far from equilibrium.
Moreover, the non-linear model for case 3, when biomass grows and substrate declines, shows growth-retarding effects at higher
substrate concentrations that are not captured by the linear substrate inhibition model. Even though linearized models make the
dynamics simpler, they may not adequately capture the effects of changing initial concentrations because they do not take into
consideration the particular parameter sensitivities. This disparity demonstrates how poorly linear models can capture biological
systems' intrinsic nonlinearity.
3. Case 1 linearized model with 50% step function on:

Figure 8: Plot representing the concentration vs time with 50% step

Figure 7: Plot representing the concentration vs time with 50% step function
function on inlet substrate concentration
on dilution rate

a) Dilution rate (dilution_rate):

Figure 7 illustrates a controlled increase in the dilution rate from the baseline 0.3 hr −1 to a higher 0.45 hr −1 after 100 hours, indicated
by the inflection points on the plots. We next analyse the reaction to this step change in biomass x 1and substrate x 2concentrations.

6
Both models show a drop in the post-step change for biomass x 1which is indicative of an increased removal rate from the reactor. The
d x1
greater size of the term −D x 1in the biomass rate of change, , mathematically demonstrates this. The Monod and Substrate
dt
Inhibition models settled at 1.46849 g/L and 1.43500 g/L, respectively, indicate a quick adjustment to a new steady state. A sensitivity
to the high substrate conditions that may prevent biomass growth is implied by the Substrate Inhibition model's greater decline.
The dilution rate introduces more substrate, which causes an initial spike in concentration for substrate x 2. However, the models
indicate that, as a result of the extra substrate's increased biomass consumption, the steady state values will eventually recover. There
are slight variations in the Substrate Inhibition model, suggesting that within the simulation period, equilibrium will almost certainly
be reached.
b) Inlet substrate concentration ( x 20 ) :
Similar step function raises the inlet substrate concentration x 20 to 6 g/L after 100 hours in the setting of Figure 8. Since there is more
substrate available for growth, biomass x 1concentrations increase as a result. Although x 20is not explicitly included in the biomass
growth calculations, its increase affects x 2in the reactor, which in turn causes an increase in x 1. Furthermore, because the system
needs time to use the extra substrate, the rate of biomass growth might not accelerate.
Once x 20 increases, the substrate concentrations x 2rise at first; but, as the biomass absorbs the excess, the system returns to its
previous steady state. Approaching a new steady state, the Substrate Inhibition model exhibits less linear behaviour, showing small
fluctuations around 0.174593 g/L. This indicates that the system is trying to reach a new equilibrium within the given time period but
is not quite successful in doing so.